Published in the Lancet

Background : Artificial intelligence (AI) has the potential to improve diagnosis. Yet people are often reluctant to trust automated systems, and some patient populations may be particularly so.

Methods: After structured interviews with patients, a randomized, blinded, factorial survey experiment placed mock patients into clinical vignettes with eight manipulated variables. A major survey firm (YouGov) provided n=2471 responses, oversampling diverse populations, with reweighting to represent the U.S. population. The primary outcome was selection of AI clinic versus human physician specialist clinic (“AI uptake”).

Results: 47.1% of Americans chose the AI clinic. In unweighted experimental contrasts, a primary care physician’s (PCP) explanation that AI has proven superior accuracy increased uptake (OR=1.48, CI 1.24-1.77, p<.001). A PCP’s nudge towards AI also increased uptake (OR=1.25, CI: 1.05-1.50, p=.013), as did reassurance that the AI clinic would listen to the patient’s unique perspectives (OR=1.27, CI: 1.07-1.52, p=.008). Disease severity (leukemia versus sleep apnea) and other manipulations did not affect AI uptake significantly. Compared to White respondents, Black respondents selected AI less often (OR=.73, CI: .55-.96, p=.023) and Native Americans selected it more often (OR: 1.37, CI: 1.01-1.87, p=.041). Populations less likely to choose AI included those who were older (OR: .99, CI: .987-.999, p=.03), politically conservative (OR: .65, CI: .52-.81, p<.001) and viewing religion as important (OR: .64, CI: .52-.77, p<.001).Andrew Keane Woods et al.

Conclusions: Many patients appear resistant to AI in diagnosis, but proof of enhanced accuracy, along with nudging by caregivers and a listening patient experience, may help increase acceptance.